Implementing traceability with EQAS The risk management of ... meeting 2016/Thelen.pdfPpt0000007...
Transcript of Implementing traceability with EQAS The risk management of ... meeting 2016/Thelen.pdfPpt0000007...
Implementing traceability
with EQAS
The risk management of trueness
Marc Thelen, SKML, Netherlands
10th anniversary of CIRME meetings, 17-18 November 2016
Risk management of trueness
1. ISO 15189:2012, 5.5.1.3Method validation, verification
– T=0; trueness and imprecision are fit for intended use
2. ISO 15189:2012, 5.6.2Internal QC
– Conditions as on T=0 are still true
– Or if not, still meet criteria of acceptance on t=0
3. ISO 15189:2012, 5.6.3External QC
– What can be wrong?
Risk of new bias, with stable IQC
Risk of new bias, with stable IQC
Risk of new bias, with stable IQC
False alarm
Goals for perfect EQAS
• Materials as intended by ISO17043:
Commutable, homogenous, stable
– Material cannot be blamed
• Value assignment in reference labs using
refererence methods
– Value assignment cannot be blamed
• Smart reporting
– You know what to do
SMART scoring of EQAS by SKML
• Specific:– Performance of lab, not of EQAS material
– Material and assigned values beyond discussion
– Support root cause analysis: method grouping
• Measurable:– Mathematical relation between statistical dispersion of results and
score
• Achievable:– if SA precision profile>Tea, then score in SA precision profile
• Realistic:– TEa based on EFLM performance goal criteria
• Time dependent– time weighed regression
Dif
fere
nce
mm
ol/
l
mmol/l
Example SMART scoring by SKML
10
Room for improvement?
Please, DO change appropriately
How good is good enough?
• EFLM second strategic conference on analytical performance specifications (APS), Milan 2014
• TFG-PSEQA“Performance specifications for EQAS”
• Chair: Graham Jones
• Members:– Stéphanie Albarède – France
– Gabriela Gutiérrez - Spain
– Mauro Panteghini - Italy
– Marc Thelen - The Netherlands
– Anne Vegard Stavelin - Norway
– Annette Thomas - UK
– Pat Twomey - UK
– Emma Ventura - Spain
Aim of the TGF-PSEQA
• Terms of Reference
– To define performance specifications for the most common measurands that should be used by EQAS organisers (for category I EQAS).
– This TFG will take into account the recommendation of the TFG-TE and TFG-DM.
• TFG-TE “Total error”
• TFG-DM “Allocation of laboratory tests to different models
for performance specifications”
• Deliverables
– A manuscript dealing with these recommendations.
Manuscript so far:
6 factor terminology PSEQA
Requirements for documenting EQA
1.EQA material commutability;
2.the method used to assign the target value; – Reference laboratory
– Certified reference materials
– Formulated (weighed in)
– statistical
3.the data set to which APS are applied;– Single measurements of single sample
– Multiple measurements of single sample
– Multiple samples
4.the applicable analytical property being assessed – Total error
– Bias
– Imprecision
6 factor terminology PSEQA (2)
5. the type of the Milan model(s) used to set the APS
– Outcome-based (Milan 1a)
– Based on clinical decision applications (Milan 1b)
– Derived from biological variation (Milan 2)
– State of the art, defined as the highest level of analytical
performance technically achievable in that moment (Milan 3).
6. the rationale for the selection of the APS
– Passable: everyone should theoretically pass
– Satisfactory: good performing laboratories should pass
– Favourable: no clinical benefit of further improvement
– Aspirational: aim to improve performance, educational.
PSEQA requirements
• All aspects communicated to (potential)
participants
• For every measurand
• In every scheme
• Possibly multiple APS for single measurand:
– Intended use !
Budget versus source
Total mu budgetAvailable mu budget for end-user
total mu budget smaller than mu
No mu budget available for end-user
IVD regulation 2016 annex 1
17.3. Information in the instructions for use
17.3.1. The instructions for use shall contain the following particulars:
•…
•XIX: The metrological traceability of values assigned to calibrators and trueness-control materials, including identification of applicable applied reference materials and/or reference measurement procedures of higher order, and information regarding maximum (self-allowed) batch to batch variation provided with relevant figuresand units of measure
•NB: this text is result of succesful lobby work by our EFLM IVD WG for 5 years ☺
•Implementation required within 5 years
How to divide the budget?
Task:
Formulate how to calculate and express the traceability and the uncertainty of value
assignment of a IVD product
Group:
ISO TC212 WG2, CEN TC140 IVD industry WG, EFLM WG accreditation and IVD
Task:
Discuss a better distribution of the total uncertainty budget between different steps in
chain
Group:
EFLM quality committee, WG accreditation and IVD
CEN TC140 IVD industry WG
laboratories IVD industry
Conclusions
• Class 1 EQC should have– Commutable samples
– Value assignment in reference methods
– Reports based multiple sample evaluation
• Class 1 EQC is needed for– Participating laboratories: local mistakes
– IVD industry: success of standardisation and harmonisation
• EFLM TFG PSEQA– Transparency of EQA APS and rationale
– Harmonisation of EQA
• New IVD regulation– Opportunity for transparency on IVD product uncertainty
– Opportunity for redistribution of total uncertainty budget